Careers at WisePort AI

WisePort AI is a pre-seed, research-led startup working toward general intelligence that can reason, plan, and act in messy, high-entropy real-world systems.

We're assembling a founding team with real ownership and direct influence on technical direction. We are looking for people who have both conceptual clarity and experience building AI systems at scale.

How We Work

✓

Early on, everyone works across boundaries. You'll occasionally step outside your core focus to help debug, prototype, or unblock the system.

✓

We optimize for research velocity: design → run → learn → adjust.

✓

We value rigorous benchmarking as our North Star for progress, and we use those gains as a springboard to solve for real-world systems.

✓

This is applied research: we explore and ship simultaneously.

✓

We value deep technical intuition, creativity and the ability to own a problem end-to-end without a roadmap.

Open Roles

The below titles describe problem spaces, not rigid boxes. If you bring deep expertise and can own hard problems, we'll find the right fit.

The Problem

We're building autonomous AI systems that reason, plan, and act. That foundation starts with training: getting models to learn the right behaviors at scale, understanding what emerges, making deliberate tradeoffs between capability and efficiency.

You'll own the research and engineering of training at this frontier. Not just "run this config." But understand: Why does this diverge? What does this scaling law tell us? How do we architect for reasoning?

What You'll Do

Design training strategy. Choose optimizers, manage hyperparameters, understand scaling dynamics.
Work directly with agent rollouts and long-horizon behaviors to understand how training choices affect real agent performance.
Research training dynamics. Why do certain approaches work at scale? What emerges? What breaks? These are research questions you'll investigate empirically.
Optimize for tradeoffs. Balance capability vs. efficiency. Speed vs. stability. Training cost vs. inference cost. Make deliberate architectural choices knowing their implications.
Collaborate across the stack. Your training strategy informs how systems are built, and system constraints shape your training choices.
Ship and iterate. Design experiments, run them, learn, adjust.

Who You Are

Core

Deep expertise in training large models (you understand optimization, scaling laws, what makes training work at scale)
Have built and shipped production systems at scale, taking ideas from papers and prototypes all the way into production
Think independently: you debug problems, question approaches, understand why things work
Comfortable with research-level thinking applied to engineering problems
Deep familiarity with distributed training primitives (FSDP, DeepSpeed, Megatron-LM) and/or low-level optimization (CUDA, Triton)

Representative experience paths (one or more)

PhD in ML/CS + 1-2 years working on production training systems
5+ years in ML engineering focused on training or optimization
3+ years at major AI labs doing this work
Strong track record going deep (published research, major open-source work, proven shipping)

These are broad experience signals, not strict requirements. If this role excites you, we'd rather talk than have you self-select out.

Mindset

You read papers and think critically about them
You're comfortable with ambiguity: we're exploring, not executing a fixed plan
You understand tradeoffs (you don't just optimize for one metric)
You take ownership and solve problems hands-on

How to Apply

Email hello@wiseport.ai with:

Resume/CV
A hard problem you solved (2-3 sentences)
- Tell us about a time training broke and how you debugged it
- OR a time you optimized something and learned unexpected lessons
Why this excites you (2 sentences)
- Why agentic AI?
Links (optional)
- Papers, GitHub, blog, anything showing your thinking

The Problem

We can't iterate fast without world-class infrastructure. Agentic systems demand more than static benchmarks: we need environments, evaluations, and tooling that reflect real behavior and failure modes.

You'll build the backbone: agentic evaluation frameworks, systems for distributed training and testing environments where we can safely experiment.

What You'll Do

Build training infrastructure: Distributed training pipelines, logging, monitoring, and experiment tracking.
Design evaluation frameworks: Agent-driven, automated, repeatable evaluation of agent behavior across success rate, correctness, robustness, controllability, and safety.
Create testing environments: Controlled spaces where we can safely test agent behavior at scale.
Optimize for speed: Design systems that enable fast iteration across training and evaluation.
Handle parallelization and scale: Design systems that scale training and evaluation efficiently.

Who You Are

Core

Strong experience building testing or evaluation infrastructure for complex systems.
Have shipped large-scale production systems.
Think independently: you see bottlenecks and know how to eliminate them.
Strong software engineering fundamentals (clean design, testing, monitoring).
Expertise in high-throughput data pipelines and secure, sandboxed execution environments (Docker, Kubernetes) for parallelized agent rollouts.
Experience with RL libraries and custom environment integrations (Gymnasium, Brax, Isaac Gym)

Nice to Have

Distributed systems or parallelization experience.
ML infrastructure experience in research environments.
Deep expertise in distributed systems and systems design.

Representative experience paths (one or more)

PhD in CS + 1-2 years building distributed systems or ML infrastructure at scale.
5+ years as a systems engineer or ML engineer focused on infrastructure and scale.
3+ years at major labs building research infrastructure at scale.
Strong track record in distributed systems, parallelization, or ML platforms.

These are broad experience signals, not strict requirements. If this role excites you, we’d rather talk than have you self-select out.

Mindset

Systems thinking: you optimize end-to-end, not just components.
Pragmatic: you value working systems over perfect ones.
Research-adjacent: you understand which evaluation questions actually matter.
You take ownership and solve problems hands-on.

How to Apply

Email hello@wiseport.ai with:

Resume/CV
A hard problem you solved (2-3 sentences)
- Tell us about infrastructure you built that enabled fast iteration.
- What was the bottleneck, and how did you fix it?
Why this excites you (2 sentences)
- Why agentic AI research infrastructure?
Links (optional)
- GitHub, blog posts, open-source work, anything showing systems thinking.

The Problem

We're building AI systems that reason and plan. That's not just an engineering problem, it's a research problem. How do models learn to reason? What mechanisms enable planning? How do these capabilities emerge and scale? Are there alternate architectures we should be looking at?

You'll explore these questions. Rigorously, empirically, and in service of building systems that live in the real-world.

This is not a purely theoretical role. You will write code and integrate your work into running systems.

What You'll Do

Own research direction in reasoning and planning. Design experiments to understand how models learn to reason and plan: what data matters, what training approaches work, and how these capabilities scale.
Understand emergence at scale. You’ve seen how models change as they get bigger, know what to look for, and bring intuition about what might emerge and what actually matters.
Collaborate across the stack. Your research informs training strategy, systems design, and RL strategy, turning insights into concrete decisions.
Think rigorously and pragmatically. You read papers critically, question approaches, and run careful experiments while still moving fast.
Ship research fast. Design experiments, run them, learn, adjust. Research velocity matters as much as rigor.

Who You Are

Core

Deep expertise in reasoning and planning in large language models.
Understand how these capabilities work at scale: you’ve seen them emerge, plateau, break, and improve.
Think rigorously about why approaches work or fail: you can ablate and debug systematically.
Have translated research insights into actual systems.
Deep expertise in search and data strategy: building synthetic data flywheels, designing reward models and scaling inference-time compute to push model capability.
Familiarity with Reinforcement Learning as it relates to long-horizon planning.

Representative experience paths (one or more)

PhD in ML/AI focused on reasoning or planning + 1-2 years working on production systems.
5+ years doing research on reasoning or planning in LLMs (industry or academia).
3+ years at major AI labs working on reasoning or planning.
Strong publication record plus proven ability to ship research into practice.

These are broad experience signals, not strict requirements. If this role excites you, we’d rather talk than have you self-select out.

Mindset

You’re genuinely curious about how reasoning and planning emerge in AI systems.
You're dissatisfied with current reasoning and planning capabilities: you see room to improve dramatically.
You’re comfortable with failure: most experiments won’t pan out, and that’s the point.
You think both rigorously (careful experiments) and pragmatically (ship fast).
You take ownership and solve problems hands-on.

How to Apply

Email hello@wiseport.ai with:

Resume/CV
A research insight you discovered (2-3 sentences)
- Tell us about an experiment or research direction that surprised you.
- What did you learn? How did you validate it?
Why this excites you (2 sentences)
- Why reasoning in agentic AI systems?
Links (optional)
- Papers, GitHub, blog, anything showing your research thinking.

The Problem

Autonomous systems need to make decisions, learn from feedback, and adapt their behavior over time. That's where Reinforcement Learning (RL) comes in. But not in a naive way: How do we design reward structures that actually work? Where does RL fail? How do we build systems that are capable, controllable and adaptable to new environments?

You'll research these questions. Understand the frontier. Push it forward.

What You'll Do

Own research direction in agentic behavior. Design how models learn to act, adapt and plan: what reward structures work, what exploration strategies matter, and how failure modes arise.
Understand where RL works and fails. You’ve seen RL systems succeed and break, know the patterns, and bring intuition about what to watch for.
Design for safety and controllability from the ground up. Build alignment into the RL approach, not as a patch.
Collaborate across the stack. Your research informs how RL integrates into training strategy and systems design.
Ship research fast. Design experiments, run them, learn, adjust. Research velocity matters as much as rigor.

Who You Are

Core

Deep expertise in reinforcement learning: algorithms, reward design, exploration, and policy learning.
Have shipped RL systems that live in the real-world: you understand the critical path from simulation to deployment and know what matters for transfer.
Strong mathematical foundation: you reason rigorously about algorithms.
Analyze rigorously which approaches work or fail, and why: you can ablate and debug systematically.
Real-world agent debugging experience; robotics background is a huge plus.
Experience with RL libraries and custom environment integration (Gymnasium, Brax, Isaac Gym): you know how to bridge algorithmic theory and real-world complexity.

Representative experience paths (one or more)

PhD in RL, Control, or Robotics + 1-2 years shipping RL systems.
5+ years doing RL work in industry or labs.
3+ years at major AI labs doing this work.

These are broad experience signals, not strict requirements. If this role excites you, we’d rather talk than have you self-select out.

Mindset

You think rigorously about algorithms and systems.
You’re practical: you understand failure modes and how to avoid them.
You’re safety-conscious: you think about misalignment and reward hacking.
You take ownership and solve problems hands-on.

How to Apply

Email hello@wiseport.ai with:

Resume/CV
A hard RL problem you solved (2-3 sentences)
- A time RL broke and how you debugged it
- OR a reward structure you designed and unexpected lessons learned
Why this excites you (2 sentences)
- Why RL in agentic AI systems?
Links (optional)
- Papers, GitHub, robotics work, anything showing your RL thinking.

FAQs

I don't have all the background listed for a role I'm interested in. Should I still apply?

If you have deep expertise in your domain and have demonstrated that expertise (through shipping, publishing, or building), apply.

I couldn't find a role that perfectly fits my background. Should I still apply?

Absolutely. We're looking for deep expertise and people who can own hard problems. If you bring skills that can move agentic AI forward: pretraining, reasoning, RL, infrastructure, alternate architectures, or otherwise, we want to hear from you.

Can I work remotely?

We're based in Gurugram (NCR), India. In-person is preferable for critical moments, but we're flexible for the right person.

What's the timeline?

We're starting conversations now. Start timing depends on role and fit.

What about compensation and equity?

We discuss salary and equity early and transparently. These are founding roles with meaningful ownership; specifics depend on experience and role scope.

We are building with people who have high technical taste and a hunger to win: those who think deeply and want real ownership over hard problems.

If that resonates, we want to talk. Reach out at hello@wiseport.ai

Careers at WisePort AI

How We Work

Open Roles

Research Engineer: Training & Optimization

The Problem

What You'll Do

Who You Are

How to Apply

ML Engineer: Agent Environments & Evals

The Problem

What You'll Do

Who You Are

How to Apply

Research Scientist: Reasoning & Planning

The Problem

What You'll Do

Who You Are

How to Apply

Research Scientist: RL & Agentic Systems

The Problem

What You'll Do

Who You Are

How to Apply

FAQs

I don't have all the background listed for a role I'm interested in. Should I still apply?

I couldn't find a role that perfectly fits my background. Should I still apply?

Can I work remotely?

What's the timeline?

What about compensation and equity?